TOA Podcast Studio: Unity’s Danny Lange on how AI is changing game development
Danny Lange views human interaction with technology as a dialogue. Lange started his work in artificial intelligence with natural language processing, but found himself early to the party. Now, after bringing his expertise to several fields — including self-driving cars and touchscreen technology — he works at Unity Technologies, the company behind the successful Unity engine.
Samsung NEXT’s Ricardo Mendez interviewed Lange, who is the vice president of AI and Machine Learning at Unity, as part of the TOA Podcast Studio series at the Tech Open Air conference in Berlin. They talked about AI, the bright side of failure, innovation culture, and algorithms.
Although Unity is primarily a game creation engine, Lange believes it as an ideal testing ground for AI applications and hopes to leverage the company’s creativity and user-first mentality to drive innovation in the AI space.
Learning from experience
Lange has helped run two startups in his career, both involved with speech recognition technology. Both startups failed because, he says, the technology wasn’t advanced enough for the solutions they were working on.
“In speech recognition, we basically came to the point where the recognizer is just not going to be good enough for the next 10 years for us to accomplish this,” he said. “We probe, we approached it from multiple sides and the line was very firm and at that point we were confident that we were too early.”
But Lange doesn’t think these failures were necessarily negative, because they added to his experience. When he moved to new areas of technology these experiences acted as “very important stepping stones” in his career, he said.
Lange believes failure can provide the experience needed for future projects and career opportunities. In his case, experience also enabled him to bring his unique way of thinking about technology to new spaces.
“When you recognize failure at one point, and you shift into another area, you bring all the experience with you,” he said. “I’m currently working on robotics. I’ve never worked on robotics before, but there are constantly things I bring with me into that new domain that allow me to attack existing, somewhat unsolvable problems by basically coming in with a whole new perspective.”
Lange is drawn to companies that embrace new ways of thinking. Unity Technologies was one such place. Driven by the need to empower creators, Lange sees a willingness to try new things and think outside of the box.
But startups and young companies aren’t the only places to find innovation. Lange thinks an innovation culture is possible anywhere. “I really have consistently gone after companies that really have that spirit to try new things and Uber’s definitely one of them,” he said. “Microsoft was one of them too. Believe it or not, IBM Research is that place, too. There’s a lot of deep-seated culture there in exploring new things.”
The future of AI
When thinking about the future of innovation in AI, Lange says there needs to be more of a focus on data. While human interaction and algorithms are crucial, he thinks the real innovation moving forward is going to come from data, not smart new algorithms.
“It’s about feeding the computer system enough data for it to figure it out,” he said. “It’s not about algorithms where everything is thought out by human designer, but it’s really about systems that are able to interact with the environment, first through simulations, and then through a real feedback loop with the physical environment that the computer learns all the things that it needs to learn to be a smart partner to us.”
For Lange, data-powered machines are essential for the evolution of AI. The game engine made by Unity, for example, is its own environment, perfect for feeding data to a learning program over and over.
“What fascinates me about the game engine,” Lange said, “is that I can create spatial scenarios where I can play out all kinds of very challenging scenarios and a computer system can, at scale, try out billions and billions of different problems and figure out how to systematically learn from that and solve those problems. That’s really how we are going to solve a lot of our challenges as humans.”
In the end, Lange wants machines to do the heavy lifting. From his point of view, his job is to make sure those AI-powered machines work well. “I really want the machines to learn to do the clever stuff,” he said, “because that’s the only point where we’re really going to be able to leverage the machines.”
Stay tuned for each new episode of the TOA Podcast Studio series on the What’s NEXT podcast feed. Co-hosted by TOA and Samsung NEXT, these sessions brought together a dozen of today’s leading minds in technology to share the human stories behind their innovations. To hear more, subscribe to What’s NEXT on Apple Podcasts, Google Play, Spotify, or via RSS on your podcast app of choice.